[Wien] Problems with velocity calculations

[Gregory Sergeev]

Dear all,
the purpose of my calculations is to get dE/dk of charge carriers for specific band number in point (x,y,z).
For this after common wien calculations I use two methods (I do calculations for Bi2Te3 in symmetric point (x,x,x)):

1. I create klist file with points with coordinates:
	(x+dx,y,z)
	(x-dx,y,z)
	(x,y+dx,z)
	(x,y-dx,z)
	(x,y,z+dx)
	(x,y,z-dx)
   Do lapw1 (and lapwso), and get E(...)'s
   Calculate dE/dk: dE/dk_x = (E(x+dx,y,z)-E(x-dx,y,z))/2dx and e.t.c.

2. I clear klist file and put there point with coordinate (x,y,z)
   Do optic (with pmat flag)

I have few problems:
1. In first method results dramatically depend on dx, for dx=1e-6 and dx=1e-7 results differ by two times... no convergence.
2. I get completely different results by this two methods.
I understand that I could lose some factor, but in first method I have symmetric vector, like (x,x,x) as expected, in second I have not symmetric result, I get vector like (x,y,z).

I have few questions:
1. What method is right (optimal)?
2. Why results are not equalt? 
3. Why I haven't convergence in first method?
4. Why in second method I receive non symmetric vector? (Even if result is in cartesian coordinates I expect vector like (0,0,x) for this point in Bi2Te3).

Hope somebody can help me or give advice.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20160629/f03bc362/attachment.html>